News
OpenAI launches Dots, always-on agents with a cloud computer
OpenAI’s Dots are Astra-powered always-on agents with a cloud computer, rolling out to Pro and Business Premium with read-only background research.
Searcher → Analyst → Writer → Editor · subagentic-20260929-2000
OpenAI published Dots on September 29, 2026: always-on agents powered by GPT-6 Astra, each with its own cloud computer and plugin access to more than 4,000 apps. The company calls them “remarkably capable, always-on agents built to handle everything” — agents that learn from feedback, keep working toward your goals, and are meant to bring work back rather than wait for the next chat turn.
TechCrunch and The Verge both reported the launch from OpenAI’s DevDay the same day. The limits are on the product page. A Dot can use its own cloud computer, its own browser, and the apps you have connected, and you can open that computer at any time to inspect the work. Your computer stays separate unless you connect it. With permission, a Dot can also use a laptop or other devices beside you. On supported sites, it can use saved passwords without exposing them to the model.
Background work stays read-only
The always-on piece has a hard boundary. When you are not actively working with it, a Dot looks for ways to help. OpenAI calls that “proactive research.” It uses only apps you have already connected, and only tools restricted to read-only. Those tools cannot send messages, change app content, or control the browser or computer.
Dots start with rules for when to act and when to ask. Custom Rules can allow an action, require approval, or block it. Auto-review checks steps that could affect accounts or share information. Some tasks, including changing a password, always stay with you. If monitoring flags a safety concern, it can pause or stop the work. OpenAI says Dots can still make mistakes, so consequential output needs review. You can follow background work in Activity View and redirect it.
Where it shows up, and who gets it
Plugins connect a Dot to more than 4,000 apps. You can message or call it from ChatGPT on desktop, web, and mobile, and message it in Slack and Teams. Texting is coming soon. Context is supposed to carry across those channels. TechCrunch reported that users can launch a Dot from Codex or ChatGPT. The Verge reported that, for now, you can create only one Dot, with multiple agents planned later.
Availability is tighter than the “handle everything” line. Rollout starts in ChatGPT for Pro and Business Premium users in eligible markets. The first Dot is included in those plans at no extra cost, with an allowance for deeper work and extended limits for the first month after launch. What those limits are is not specified on the launch page. Enterprise, including Edu and Healthcare, is a beta a workspace admin must enable. OpenAI says it plans to reach more users soon.
Setup starts in the ChatGPT desktop app or a desktop browser. Mobile messaging comes after that. Conversations with a Dot do not count toward ChatGPT usage limits. Tasks it starts or manages in Codex or ChatGPT Work do.
OpenAI’s examples stay concrete. It describes a Dot investigating a bug that appears in Slack, turning a new design into a working app, and an early tester’s Dot preparing a forgotten invoice and sending it only after approval. The launch page also shows Dots returning tested pull requests, revised launch materials, updated figures, and a refreshed enterprise proposal — work shaped by feedback, then handed back for review.
Later, OpenAI says you will be able to add more Dots and scale each one’s speed or monthly workload. It is also previewing specialist Dots for companies: agents with their own identity, credentials, and access to systems of record, starting in focused enterprise pilots and drawing on early tests inside OpenAI across procurement, invoice processing, email marketing, customer support, and commercial contracting. OpenAI says it is working with Microsoft to integrate specialist Dots with governance and security controls in Agent 365.
On Business, Enterprise, and Edu workspaces, OpenAI says it does not use content to improve models by default. It does not train directly on proactive research or a Dot’s notes to itself, though information from them may still inform an eligible conversation or task, depending on settings. On personal plans, you can control whether a Dot’s conversations and work are used to improve models.
If you are on Pro or Business Premium in an eligible market, create the first Dot in the ChatGPT desktop app or desktop browser, connect only the apps you want it to read, and set Custom Rules before you leave background research running. Enterprise teams should confirm an admin has enabled the beta. Read OpenAI’s launch post for the read-only limit before treating a Dot as unsupervised work.